British fintech is usually measured by the apps in people’s pockets, by account numbers, card designs and the market share taken from the high street. The part of the sector selling hardest to overseas buyers has no consumer following at all, and no marketing budget to build one.
Core banking and banking-as-a-service platforms sit beneath every current account, savings product and payment instruction. A cluster of British companies now builds and licenses that layer to banks around the world, on contracts measured in decades. The financial year to 31 March 2026 was the first period in which the category produced numbers that survive close reading.
Timing has helped. A Bank of England staff working paper published in August 2026 examined two decades of outsourcing agreements and found cloud spending associated with lower operating costs and higher deposits. The commercial argument for the entire sector has arrived from the central bank.
The £70m signal
Engine by Starling recorded revenue of £10.9m for the year to 31 March 2026, growth of 25% on the previous period, while its client base doubled to four. The subsidiary employs around 300 people, and the group invested £19.7m in it during the year to expand engineering capacity and international sales.
The recognised revenue understates the position. Engine closed the year with £70m of committed annual recurring revenue and has set a target above £100m, following a ten-year agreement with Tangerine, its first client in North America, and a subsequent deal with SBS Bank in New Zealand. Software revenue of this type is recognised as customers migrate, so the committed figure leads the reported one by several years.
That distinction matters for how the parent is valued. Starling reported group revenue of £887.4m and pre-tax profit of £217.1m for the year, with 6.2 million platform accounts, £12.7bn of customer deposits and a capital surplus above £525m as at May 2026. Public markets price recurring software revenue on entirely different terms to bank earnings, and the group carries both inside one balance sheet ahead of an eventual listing.
Evidence from the regulator
Bank of England Staff Working Paper No. 1,199 draws on a proprietary dataset of 3,443 material outsourcing contracts from 90 UK domestic banks covering the past two decades. Of these, 457 involve cloud service arrangements with external providers, accounting for annual spending of £1.57bn across 72 banks. A 10% increase in cloud spending is associated with roughly a 1.2% reduction in non-interest expenses and a 1.0% reduction in staff expenses.
The paper’s two findings sit at different levels and are worth separating. The reduced-form cost effects concentrate among large institutions and building societies, while the structural model of deposit competition shows the demand-side benefits falling substantially in favour of small and medium banks and building societies. A counterfactual in which cloud outsourcing had been restricted before widespread adoption implies higher market concentration and reduced market shares for smaller institutions.
Stated plainly, the central bank has concluded that cloud outsourcing has partly lowered the technological barriers to competition in British banking. Every platform vendor in the country has been making a version of that argument for a decade. Hearing it from Threadneedle Street, supported by contract-level data no supplier could assemble, changes who is willing to act on it.
Profit finally arrives
Thought Machine passed $100m of total revenue for the financial year ending December 2025, a 57% increase, with annual recurring revenue crossing the same threshold in the second quarter of 2026. The company turned free cash flow positive during the second half of 2025 and cut pre-tax losses from nearly £70m to around £12m. Its platform now runs at 68 banks in over 30 countries, including 18 of the world’s largest institutions, and the London headquarters plans to add more than 100 engineers during 2026.
Chief executive and founder Paul Taylor has framed the milestone as proof that the largest banks are deploying cloud-native technology “at scale for full bank migrations”, not confining it to new digital brands. 10x Banking makes a similar point from a different starting position. Founded in 2016 by Antony Jenkins, it became EBITDA-positive in the fourth quarter of 2025, lifted annual recurring revenue by more than 30% in the twelve months to May 2026, passed 10 million live customer accounts and onboarded more than ten new financial institutions before raising £40m in August 2026.
Raising capital after profitability, and using it for sales capacity, inverts the usual sequence in enterprise software. It also tells prospective bank customers something about vendor durability, which is the first question any migration committee asks.
The wider infrastructure layer
ClearBank shows how the model performs when the operator holds a licence of its own. Group normalised revenue rose 34% to £121.6m for 2025, with fee-based income up 51%, and the UK bank delivered a third consecutive profitable year with pre-tax profit of £12.2m, up 53%. Customer balances increased 65% to £17.8bn, payment volumes rose 57% to 262 million transactions, and the platform now supports more than 17 million accounts, up from 13 million. The UK entity has since secured an investment grade credit rating of BBB minus.
Its European arm onboarded 21 clients during its first full year of operations, taking the total to 28, opened a branch in Paris and had passported into 21 European Union markets by early 2026. Form3 occupies the payments equivalent of the same position, running cloud-native account-to-account infrastructure for banks and payment institutions, and secured strategic investment and a debt facility during 2025 to fund expansion into the United States. Griffin, meanwhile, holds a full British banking licence granted in March 2024, having applied in May 2022, which allows it to combine deposit-taking with the software layer other providers rent from partner banks.
Four different commercial structures, one export. Each sells British-built financial infrastructure to institutions that could have bought from anywhere.
Fifteen year decisions
Choice of core platform is among the longest-dated decisions a bank board takes. Migration timelines run for years, the systems then operate for a decade or more, and the switching costs after implementation are severe enough that most institutions only reconsider under duress. Vendors are effectively underwriting a fifteen-year relationship at the point of signature.
Revenue of that character is unusually high quality once won. It is contracted, recurring, largely independent of interest rates and it grows as the customer’s own account base grows, which is why Engine’s committed figure sits seven times above its recognised revenue. The same duration explains the sector’s long unprofitable adolescence. Sales cycles measured in years consume capital before they produce cash, and only now is the first cohort of British platforms reaching the point where earlier wins pay for current growth.
What Britain gains
The economic case is straightforward. Services exports of this type carry high margins, employ engineers at British salaries, generate revenue in dollars and euros, and cannot easily be relocated once the intellectual property and the client relationships are established here. Thought Machine alone serves clients in more than 30 countries from a London base, and 10x Banking supports institutions across Australia, New Zealand, South Africa and the United Kingdom.
There is a second dividend that is easier to miss. The Bank of England’s finding that smaller institutions capture the larger share of demand-side benefits from cloud adoption suggests domestic competition improves as the technology spreads. Building societies and challenger banks gain access to infrastructure that was previously the preserve of institutions able to fund their own data centres. British platforms are the most likely suppliers of it, which means the export industry and the domestic competition agenda now pull in the same direction.
Where the opportunity sits
Reading the sector accurately requires attention to a small number of figures. Committed annual recurring revenue matters more than reported revenue, because it captures signed business ahead of migration. Contract duration and client concentration determine how much of that committed revenue is genuinely durable. Free cash flow, EBITDA position and the size of prior-year losses indicate whether a platform can fund migrations without returning to the market on unfavourable terms.
Exposure to the category remains largely private, through venture and growth rounds, structured credit facilities of the type 10x Banking used, secondary transactions in employee shares, and eventually a Starling listing in which the software arm may carry a meaningful share of the valuation. The risks are equally identifiable. Migration programmes run late, a handful of hyperscale cloud providers underpin the entire category, and the Bank of England has been explicit that boards remain accountable for dependencies they do not own. The counterweight is that Britain currently holds a lead in a market every bank in the world must eventually buy from.